how confident can teams be that the autogenerated tests will catch the weird edge cases, or adapt when UI or flows change mid-sprint? | discoverkit | discoverkit